The candidate's education and research have focused on environmental and occupational epidemiology, with particular interest in gene-environment interactions and modeling biologic pathways. The candidate's research in an occupational cohort study led her to observe that, studies that include workers hired prior to the start of follow-up over-sample healthier workers, which exaggerates the healthy worker survivor effect (HWSE) and induces a downward bias. This proposed training will allow her to gain skills to develop epidemiologic methods to address this and other occupational methods issues as well as enhance her potential to develop into an independent investigator focused on occupational research. Mentors and coinvestigators for this proposal have expertise in epidemiologic and quantitative methods, occupational epidemiology, and modeling occupational exposures. The training involves gaining experience in causal modeling, designing biologic models that account for an immunologic response to exposure, and developing and testing novel methods to minimize HWSE bias. The training will also provide opportunities for her to begin to establish an independent research program. The study population for the proposed work is the Shanghai female textile workers study, a case-cohort study of 3184 controls and 628 cases of lung cancer. Previously, an inverse association between lung cancer and endotoxin was reported. This is biologically plausible given that endotoxin may lead to immunomodulation, which itself may be anticarcinogenic. However, the study population includes only subjects hired prior to the start of follow-up, which may partially explain this inverse association. The specific aims are: 1) to estimate the exposure-response for endotoxin and lung cancer accounting for immune response modifiers (e.g., cessation of exposure and hormonal changes);2) to determine how much bias is introduced from the inclusion of subjects hired much earlier than the start of follow-up;and 3) to develop and compare methods for controlling for this form of HWSE bias. This research directly addresses a common problem in occupational health research - how to handle the exaggerated HWSE bias when subjects were hired prior to the start of follow-up. In addition, advanced training in developing epidemiologic methods and modeling biologic mechanisms will allow the candidate to develop into an independent investigator.